Examples with Parameter ai.djl.nn.Parameter used on opensource projects

Example 1 with Parameter

use of ai.djl.nn.Parameter in project djl by deepjavalibrary.

the class BlockCoreTest method testEncode.

private void testEncode(NDManager manager, Block block) throws IOException, MalformedModelException {
    PairList<String, Parameter> original = block.getParameters();
    File temp = File.createTempFile("block", ".param");
    DataOutputStream os = new DataOutputStream(Files.newOutputStream(temp.toPath()));
    block.saveParameters(os);
    block.loadParameters(manager, new DataInputStream(Files.newInputStream(temp.toPath())));
    Files.delete(temp.toPath());
    PairList<String, Parameter> loaded = block.getParameters();
    int bound = original.size();
    for (int idx = 0; idx < bound; idx++) {
        Assert.assertEquals(original.valueAt(idx), loaded.valueAt(idx));
    }
}

Example 2 with Parameter

use of ai.djl.nn.Parameter in project djl by deepjavalibrary.

the class GRU method forwardInternal.

/**
 * {@inheritDoc}
 */
@Override
protected NDList forwardInternal(ParameterStore parameterStore, NDList inputs, boolean training, PairList<String, Object> params) {
    NDArrayEx ex = inputs.head().getNDArrayInternal();
    Device device = inputs.head().getDevice();
    NDList gruParams = new NDList();
    for (Parameter parameter : parameters.values()) {
        gruParams.add(parameterStore.getValue(parameter, device, training));
    }
    NDArray input = inputs.head();
    if (inputs.size() == 1) {
        int batchIndex = batchFirst ? 0 : 1;
        inputs.add(input.getManager().zeros(new Shape((long) numLayers * getNumDirections(), input.size(batchIndex), stateSize)));
    }
    NDList outputs = ex.gru(input, inputs.get(1), gruParams, hasBiases, numLayers, dropRate, training, bidirectional, batchFirst);
    if (returnState) {
        return outputs;
    }
    outputs.stream().skip(1).forEach(NDArray::close);
    return new NDList(outputs.get(0));
}

Example 3 with Parameter

use of ai.djl.nn.Parameter in project djl by deepjavalibrary.

the class LSTM method forwardInternal.

/**
 * {@inheritDoc}
 */
@Override
protected NDList forwardInternal(ParameterStore parameterStore, NDList inputs, boolean training, PairList<String, Object> params) {
    NDArrayEx ex = inputs.head().getNDArrayInternal();
    Device device = inputs.head().getDevice();
    NDList rnnParams = new NDList();
    for (Parameter parameter : parameters.values()) {
        rnnParams.add(parameterStore.getValue(parameter, device, training));
    }
    NDArray input = inputs.head();
    if (inputs.size() == 1) {
        int batchIndex = batchFirst ? 0 : 1;
        Shape stateShape = new Shape((long) numLayers * getNumDirections(), input.size(batchIndex), stateSize);
        // hidden state
        inputs.add(input.getManager().zeros(stateShape));
        // cell
        inputs.add(input.getManager().zeros(stateShape));
    }
    NDList outputs = ex.lstm(input, new NDList(inputs.get(1), inputs.get(2)), rnnParams, hasBiases, numLayers, dropRate, training, bidirectional, batchFirst);
    if (returnState) {
        return outputs;
    }
    outputs.stream().skip(1).forEach(NDArray::close);
    return new NDList(outputs.get(0));
}

Example 4 with Parameter

use of ai.djl.nn.Parameter in project djl by deepjavalibrary.

the class RNN method forwardInternal.

/**
 * {@inheritDoc}
 */
@Override
protected NDList forwardInternal(ParameterStore parameterStore, NDList inputs, boolean training, PairList<String, Object> params) {
    NDArrayEx ex = inputs.head().getNDArrayInternal();
    Device device = inputs.head().getDevice();
    NDList rnnParams = new NDList();
    for (Parameter parameter : parameters.values()) {
        rnnParams.add(parameterStore.getValue(parameter, device, training));
    }
    NDArray input = inputs.head();
    if (inputs.size() == 1) {
        int batchIndex = batchFirst ? 0 : 1;
        inputs.add(input.getManager().zeros(new Shape((long) numLayers * getNumDirections(), input.size(batchIndex), stateSize)));
    }
    NDList outputs = ex.rnn(input, inputs.get(1), rnnParams, hasBiases, numLayers, activation, dropRate, training, bidirectional, batchFirst);
    if (returnState) {
        return outputs;
    }
    outputs.stream().skip(1).forEach(NDArray::close);
    return new NDList(outputs.get(0));
}

Example 5 with Parameter

use of ai.djl.nn.Parameter in project djl by deepjavalibrary.

the class RecurrentBlock method prepare.

/**
 * {@inheritDoc}
 */
@Override
public void prepare(Shape[] inputs) {
    Shape inputShape = inputs[0];
    ParameterList parameters = getDirectParameters();
    for (Pair<String, Parameter> pair : parameters) {
        String name = pair.getKey();
        Parameter parameter = pair.getValue();
        int layer = Integer.parseInt(name.split("_")[1]);
        long inputSize = inputShape.get(2);
        if (layer > 0) {
            inputSize = stateSize * getNumDirections();
        }
        if (name.contains("BIAS")) {
            parameter.setShape(new Shape(gates * stateSize));
        } else if (name.contains("i2h")) {
            parameter.setShape(new Shape(gates * stateSize, inputSize));
        } else if (name.contains("h2h")) {
            parameter.setShape(new Shape(gates * stateSize, stateSize));
        } else {
            throw new IllegalArgumentException("Invalid parameter name");
        }
    }
}